Iterative reconstruction of Fourier-rebinned PET data using sinogram blurring function estimated from point source...

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Image Reconstruction

Background:

• Accurate system modeling is crucial for high-quality Positron Emission Tomography (PET) image reconstruction.
• Iterative reconstruction algorithms benefit from factoring the system model into geometric projection and sinogram blurring functions.
• Fourier rebinning of 3D PET data into 2D sinograms allows for faster reconstruction using 2D algorithms.

Purpose of the Study:

• To develop a method for estimating the sinogram blurring function specifically for Fourier-rebinned PET data.
• To improve the accuracy of the system model used in iterative image reconstruction.
• To enhance the quality of reconstructed PET images by accurately modeling blurring effects.

Main Methods:

• Extended a previous point source-based method to estimate sinogram blurring functions for Fourier-rebinned data.
• Treated the sinogram blurring function as separable into transaxial and axial components.
• Estimated a radially and angularly variant 2D blurring function for transaxial blurring and a space-variant 1D kernel for axial blurring.
• Incorporated the estimated blurring function into a 2D maximum a posteriori (MAP) reconstruction algorithm.

Main Results:

• Validated the method using physical phantom experiments on the microPET II scanner.
• Demonstrated improved image contrast and spatial resolution compared to reconstructions without a blurring model or with a Monte Carlo-based model.
• Observed no increase in reconstruction time as the blurring component computation was negligible.

Conclusions:

• The proposed method effectively estimates sinogram blurring matrices for Fourier-rebinned PET data.
• This estimation significantly improves the contrast and spatial resolution of reconstructed PET images.
• The method is applicable to both human and animal PET scanners.